Pseudo‐Three‐Dimensional Magnetotelluric Inversion Using Nonlinear Conjugate Gradients

Abstract

Abstract We propose a nonlinear conjugate gradient algorithm for pseudo‐3‐D MT inversion. The conjugate gradient method is selected as the inversion kernel. When calculating sensitivities matrix, instead of computing 3‐D sensitivities, we compute 1‐D sensitivities. Since the sensitivity elements of non‐stations are zero, which should be non‐negligible values, we adopt a method of approximation. We update sensitivity matrix by the quasi‐Newton method after the first inversion. Pseudo‐3‐D MT inversion can save computational time greatly. Inversion results of a synthetic data set and real data set show that the pseudo‐3‐D MT inversion is reliable and feasible.